China Enhances AI Safety Frameworks to Prevent Risks

On September 15, 2023, China is advancing its safety frameworks for artificial intelligence (AI) to mitigate risks associated with AI systems that could operate independently and potentially challenge human control. Since 2024, Beijing has been integrating these risks into its safety protocols, focusing on AI agents capable of planning and executing tasks autonomously.

The approach involves developer responsibilities, government oversight, security assessments, and external testing. A policy issued in May by China's cyberspace regulator, economic planner, and industry ministry identifies "operational loss of control" as a significant security risk.

Developers must enhance their ability to detect and manage improper behaviors, while users must be informed of AI agents' autonomous decisions. Authorities are drafting a mandatory national standard for AI agent safety, which has not yet been established.

Concerns were raised by Wang Lihong, a senior cyberspace regulatory official, regarding advanced models bypassing safety measures and affecting real-world systems.